Show the code
pacman::p_load(jsonlite, tidygraph, ggraph,
visNetwork, graphlayouts, ggforce,
skimr, tidytext, tidyverse, extrafont, knitr, ggtext)Take-home Exercise 3
1. Overview
This exercise aims to use appropriate static and interactive statistical graphics methods to help FishEye identify companies that may be engaged in illegal fishing.
The original dataset was originated from Mini Challenge 3 of Vast Challenge 2023.
There is one file downloaded: MC3.json.
This exercise aims to answer Q1 of the challenge:
- Use visual analytics to identify anomalies in the business groups present in the knowledge graph. Limit your response to 400 words and 5 images.
2. Getting Started
2.1 Installing and launching R packages
The code chunk below will be used to install and load the necessary R packages to meet the data preparation, data wrangling, data analysis and visualisation needs.
2.2 Importing json file by using jsonlite package
Show the code
mc3_data <- fromJSON("data/MC3.json")2.3 Data Preparation
2.3.1 Extracting edges
The code chunk below will be used to extract the links data.frame of mc3_data and save it as a tibble data.frame called mc3_edges.
Show the code
mc3_edges <- as_tibble(mc3_data$links) %>%
distinct() %>%
mutate(source = as.character(source),
target = as.character(target),
type = as.character(type)) %>%
group_by(source, target, type) %>%
summarise(weights = n()) %>%
filter(source!=target) %>%
ungroup()2.3.2 Extracting nodes
The code chunk below will be used to extract the nodes data.frame of mc3_data and save it as a tibble data.frame called mc3_nodes.
Show the code
mc3_nodes <- as_tibble(mc3_data$nodes) %>%
mutate(country = as.character(country),
id = as.character(id),
product_services = as.character(product_services),
revenue_omu = as.numeric(as.character(revenue_omu)),
type = as.character(type)) %>%
select(id, country, type, revenue_omu, product_services)3. Initial Data Exploration
3.1 Exploring the edges data frame
In the code chunk below, skim() of skimr package is used to display the summary statistics of mc3_edges tibble data frame.
Show the code
skim(mc3_edges)| Name | mc3_edges |
| Number of rows | 24036 |
| Number of columns | 4 |
| _______________________ | |
| Column type frequency: | |
| character | 3 |
| numeric | 1 |
| ________________________ | |
| Group variables | None |
Variable type: character
| skim_variable | n_missing | complete_rate | min | max | empty | n_unique | whitespace |
|---|---|---|---|---|---|---|---|
| source | 0 | 1 | 6 | 700 | 0 | 12856 | 0 |
| target | 0 | 1 | 6 | 28 | 0 | 21265 | 0 |
| type | 0 | 1 | 16 | 16 | 0 | 2 | 0 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| weights | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 | ▁▁▇▁▁ |
The report above reveals that there is not missing values in all fields.
In the code chunk below, datatable() of DT package is used to display mc3_edges tibble data frame as an interactive table on the html document.
Show the code
DT::datatable(mc3_edges)A plot below shows the distribution of variable type in the mc3_edges data table. This variable only consists of Beneficial Owner and Company Contacts. There are much more Beneficial Owners than Company Contacts.
Show the code
ggplot(data = mc3_edges,
aes(x = type)) +
geom_bar() +
labs(title = "Distribution of type in mc3_edges data table")
3.2 Initial Network Visualisation and Analysis
3.2.1 Building network model with tidygraph
Show the code
id1 <- mc3_edges %>%
select(source) %>%
rename(id = source)
id2 <- mc3_edges %>%
select(target) %>%
rename(id = target)
mc3_nodes1 <- rbind(id1, id2) %>%
distinct() %>%
left_join(mc3_nodes,
unmatched = "drop")
mc3_graph <- tbl_graph(nodes = mc3_nodes1,
edges = mc3_edges,
directed = FALSE) %>%
mutate(betweenness_centrality = centrality_betweenness(),
closeness_centrality = centrality_closeness())
mc3_graph %>%
filter(betweenness_centrality >= 100000) %>%
ggraph(layout = "fr") +
geom_edge_link(aes(alpha=0.5)) +
geom_node_point(aes(
size = betweenness_centrality,
colors = "lightblue",
alpha = 0.5)) +
scale_size_continuous(range=c(1,10))+
theme_graph() +
labs(title = "Network model of mc3 data")
3.3 Exploring the nodes data frame
In the code chunk below, skim() of skimr package is used to display the summary statistics of mc3_nodes tibble data frame.
Show the code
skim(mc3_nodes)| Name | mc3_nodes |
| Number of rows | 27622 |
| Number of columns | 5 |
| _______________________ | |
| Column type frequency: | |
| character | 4 |
| numeric | 1 |
| ________________________ | |
| Group variables | None |
Variable type: character
| skim_variable | n_missing | complete_rate | min | max | empty | n_unique | whitespace |
|---|---|---|---|---|---|---|---|
| id | 0 | 1 | 6 | 64 | 0 | 22929 | 0 |
| country | 0 | 1 | 2 | 15 | 0 | 100 | 0 |
| type | 0 | 1 | 7 | 16 | 0 | 3 | 0 |
| product_services | 0 | 1 | 4 | 1737 | 0 | 3244 | 0 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| revenue_omu | 21515 | 0.22 | 1822155 | 18184433 | 3652.23 | 7676.36 | 16210.68 | 48327.66 | 310612303 | ▇▁▁▁▁ |
The report above reveals that there is no missing values in all fields.
In the code chunk below, datatable() of DT package is used to display mc3_nodes tibble data frame as an interactive table on the html document.
Show the code
DT::datatable(mc3_nodes)A plot below shows the distribution of variable type in the mc3_nodes data table. This variable consists of Beneficial Owner, Company and Company Contacts. There are more Beneficial Owners than Companies than Company Contacts.
Show the code
ggplot(data = mc3_nodes,
aes(x = type)) +
geom_bar() +
labs(title = "Distribution of type in mc3_nodes data table")
3.4 Text Sensing with tidytext
This section performs basic text sensing using appropriate functions of tidytext package.
3.4.1 Simple word count
This code counts the number of words related to fishing e.g. “fish”, “fishes”, “seafood”, “fishing”, etc. in the product_services column. Before that, we make the characters in product_services all lower case for ease of searching.
Show the code
mc3_nodes <- mc3_nodes %>%
mutate_at(.vars = vars(product_services),
.funs = funs(tolower))
mc3_nodesfish <- mc3_nodes %>%
mutate(n_fish = str_count(product_services, "fish|fishes|seafood|fishing|food|salmon|tuna|sea|seafoods")) A plot of the distribution of nodes with fishing related descriptions in their product services is shown below.
Show the code
ggplot(data = mc3_nodesfish,
aes(x = n_fish)) +
geom_bar() +
coord_cartesian(xlim =c(0, 25)) +
labs(title = "Distribution of ids related to fishing activities", x = "Number of fishing related words", y = "Number of ids")
3.4.2 Tokenisation
The word tokenisation have different meaning in different scientific domains. In text sensing, tokenisation is the process of breaking up a given text into units called tokens. Tokens can be individual words, phrases or even whole sentences. In the process of tokenisation, some characters like punctuation marks may be discarded. The tokens usually become the input for the processes like parsing and text mining.
In the code chunk below, unnest_token() of tidytext is used to split text in product_services field into words.
Show the code
token_nodes <- mc3_nodes %>%
unnest_tokens(word,
product_services)The two basic arguments to unnest_tokens() used here are column names. First we have the output column name that will be created as the text is unnested into it (word, in this case), and then the input column that the text comes from (product_services, in this case).
Next, we visualise the words extracted by using the code chunk below.
Show the code
token_nodes %>%
count(word, sort = TRUE) %>%
top_n(15) %>%
mutate(word = reorder(word, n)) %>%
ggplot(aes(x = word, y = n)) +
geom_col() +
xlab(NULL) +
coord_flip() +
labs(x = "Count",
y = "Unique words",
title = "Count of unique words found in product_services field")
The bar chart reveals that the unique words contains some words that may not be useful to use e.g. “and” and “of”. We want to remove these words from your analysis as they are fillers used to compose a sentence.
3.4.3 Removing stopwords
Use tidytext package that has a function called stop_words that will help to clean up stop words.
Show the code
stopwords_removed <- token_nodes %>%
anti_join(stop_words)
stopwords_removed %>%
count(word, sort = TRUE) %>%
top_n(15) %>%
mutate(word = reorder(word, n)) %>%
ggplot(aes(x = word, y = n)) +
geom_col() +
xlab(NULL) +
coord_flip() +
labs(x = "Count",
y = "Unique words",
title = "Count of unique words found in product_services field")
3.4.4 Recode words “character”, “0”, and “unknown” in product_services field to NA
To make the data more clean and meaningful, we recode the words “character”, “0”, “unknown” and “related” in product_services field to NA. It is also found that there are words “characters” and characterization”. These are recoded to NA as well.
Show the code
stopwords_removed$word[grepl("character|characters|characterization|0|unknown|related", stopwords_removed$word)]<-"NA"4. Identify anomalies in the business groups
4.1 Data preparation
4.1.1 Unlist source column of mc3_edges data table
We observed there are lists in the source column of mc3_edges data table. To clean up the list, we first segregate out rows with lists from the mc3_edges data table into a separate mc3_edges_wlist data table.
Show the code
mc3_edges_nolist <- mc3_edges %>%
filter(!grepl('c(', source, fixed = TRUE))
mc3_edges_wlist <- mc3_edges %>%
filter(grepl('c(', source, fixed = TRUE))As the source column is in chr format, we unlist the lists in source column of mc3_edges_wlist data table by removing characters “c(” and “)” and split the elements by “,” using str R package. Then, we remove any duplicate names using lapply. Next, we use the unnest_longer function to separate the elements into new rows.
Show the code
mc3_edges_wlist$source <- str_replace_all(mc3_edges_wlist$source, "c\\(|\\)|\"", "")
mc3_edges_wlist$source <- str_split(mc3_edges_wlist$source, ", ")
mc3_edges_wlist$source <- lapply(mc3_edges_wlist$source, unique)
mc3_edges_wlist <- mc3_edges_wlist %>%
unnest_longer(source) %>%
distinct()Lastly, we merge back the rows with the original mc3_edges data table (less the list) to form the cleaned edges data table.
Show the code
mc3_edges_cleaned <- rbind(mc3_edges_nolist, mc3_edges_wlist) %>%
distinct()4.1.2 Create new mc3_nodes data table
Now that we have a cleaned mc3 edges data table, we will use this to create a new nodes data table to ensure all sources and targets are captured in the nodes data table to facilitate accurate development and analysis of network graphs later.
We want to extract the source and target from the mc3_edges_cleaned data table and left join stopwords_removed nodes data table to form the new nodes table.
We first take a look at the stopwords_removed nodes data table. We observed there are duplicate ids, some of which having same id but different country. We want to combine those duplicate ids to remove duplication. This can be done by grouping by id and type and using the summarise function to concatenate the country and word. For revenue_omu, we take the median value as taking the sum may not be comparative to other single id nodes. Then, we use str_split function to split the characters by ” , ” and then use lapply to ensure no duplicates in each field of country column.
Show the code
mc3_nodesclean <- stopwords_removed %>%
group_by(id, type) %>%
summarise(country = paste(country, collapse = " , "), revenue_omu = median(revenue_omu), word = paste(word, collapse = " , ")) %>%
ungroup()
mc3_nodesclean$country <- str_split(mc3_nodesclean$country, " , ")
mc3_nodesclean$country <- lapply(mc3_nodesclean$country, unique)
mc3_nodesclean <- mc3_nodesclean %>%
select(id, country, revenue_omu, word)Then, we look at the mc3_nodes_cleaned data table. We observed that the source in mc3_nodes_cleaned data table are all companies while the target are people’s names, which are aligned to type that comprises Beneficial Owner and Company Contacts. As such, we can safely assume that type belongs to target.
We extract the source (and create a new column type and name it as “Company”) and also extract target from the mc3_edges_cleaned data table and left join mc3_nodesclean data table to form the new nodes table.
Show the code
id3 <- mc3_edges_cleaned %>%
select(source) %>%
rename(id = source) %>%
mutate(type = "Company")
id4 <- mc3_edges_cleaned %>%
select(target, type) %>%
rename(id = target)
mc3_nodes_cleaned <- rbind(id3, id4) %>%
distinct() %>%
left_join(mc3_nodesclean, by=c("id" = "id"),
unmatched = "drop")
mc3_graphcleaned <- tbl_graph(nodes = mc3_nodes_cleaned,
edges = mc3_edges_cleaned,
directed = FALSE) %>%
mutate(betweenness_centrality = centrality_betweenness(),
closeness_centrality = centrality_closeness())We are now ready to analyse the following relationships:
Company (source) and Beneficial Owner (target)
Company (source) and Company Contacts (target)
4.2 Visualization of distribution of Company ownership by Beneficial Owner
We want to check the frequency distribution of Company ownership by Beneficial Owner.
First, select edges where source type = Company and target type = Beneficial Owner. Count number of duplicated target i.e. Beneficial Owner to find out how many companies are owned by the Beneficial Owner. Then plot the distribution.
Show the code
mc3_edges_cb <- mc3_edges_cleaned %>%
filter(type == "Beneficial Owner") %>%
add_count(target) #This adds a column (n) to the table indicating the number of companies linked to the target.
ggplot(data = mc3_edges_cb,
aes(x = n)) +
geom_bar() +
labs(title = "Distribution of Number of Companies Owned by Beneficial Owner", x = "Number of Companies Owned", y = "Number of Beneficial Owners")
Show the code
kable(head(mc3_edges_cb %>%
arrange(desc(n)) %>%
as_tibble(), 10))| source | target | type | weights | n |
|---|---|---|---|---|
| Adams Group | John Smith | Beneficial Owner | 1 | 11 |
| Faroe Islands Company World | John Smith | Beneficial Owner | 1 | 11 |
| Guzman-Chang | John Smith | Beneficial Owner | 1 | 11 |
| Peterson PLC | John Smith | Beneficial Owner | 1 | 11 |
| Ryan-Curry | John Smith | Beneficial Owner | 1 | 11 |
| The Salted Pearl Inc Pelican | John Smith | Beneficial Owner | 1 | 11 |
| hǎi zhé Herring Incorporated Logistics | John Smith | Beneficial Owner | 1 | 11 |
| Beachcombers Nautical Plc Carriers | John Smith | Beneficial Owner | 1 | 11 |
| SeaSplash Foods Corporation Freight | John Smith | Beneficial Owner | 1 | 11 |
| Kerala Market Oyj Freight | John Smith | Beneficial Owner | 1 | 11 |
4.3 Visualization of relationship between Company and Beneficial Owner using network graph
4.3.1 Analysis of Beneficial Owners with more than 4 companies
First, we form new nodes data table by using source and target of the edges data table. This is to ensure that the nodes in nodes data tables include all the source and target values. Group_component <10 is used to identify the prominent communities.
Show the code
mc3_edges_cb4 <- mc3_edges_cb %>%
filter(n>4)
id7 <- mc3_edges_cb4 %>%
select(source) %>%
rename(id = source)
id8 <- mc3_edges_cb4 %>%
select(target) %>%
rename(id = target)
mc3_nodes_cb4 <- rbind(id7, id8) %>%
distinct() %>%
left_join(mc3_nodes_cleaned, by=c("id" = "id"),
unmatched = "drop") %>%
filter(type %in% c("Company", "Beneficial Owner"))
mc3_graph_cb4 <- tbl_graph(nodes = mc3_nodes_cb4,
edges = mc3_edges_cb4,
directed = FALSE)
mc3_graph_cb4 <- mc3_graph_cb4 %>%
activate("nodes") %>%
mutate(group = group_components()) %>%
filter(group < 10)
edges_cb4_df <- mc3_graph_cb4 %>%
activate(edges) %>%
as_tibble()
nodes_cb4_df <- mc3_graph_cb4 %>%
activate(nodes) %>%
as_tibble() %>%
rename(label = id) %>%
mutate(id=row_number()) %>%
select(id, label, group)Show the code
visNetwork(nodes_cb4_df,
edges_cb4_df) %>%
visIgraphLayout(layout = "layout_with_fr") %>%
visOptions(highlightNearest = TRUE,
nodesIdSelection = TRUE) %>%
visEdges(smooth = list(enabled = TRUE,
type = "curvedCW")) %>%
visLegend() %>%
visLayout(randomSeed = 123)4.3.2 Analysis of betweenness centrality between Company and Beneficial Owner (tidygraph)
Centrality Betweenness is a way of detecting the amount of influence a node has over the flow of information in a graph. It finds nodes that serve as a bridge from one part of a graph to another and measures the shortest paths between all pairs of nodes in a graph. A node with higher betweenness centrality would have more control over the network.
First, we form new nodes data table by using source and target of the edges data table. This is to ensure that the nodes in nodes data tables include all the source and target values.
Show the code
id5 <- mc3_edges_cb %>%
select(source) %>%
rename(id = source)
id6 <- mc3_edges_cb %>%
select(target) %>%
rename(id = target)
mc3_nodes_cb <- rbind(id5, id6) %>%
distinct() %>%
left_join(mc3_nodes_cleaned, by=c("id" = "id"),
unmatched = "drop") %>%
filter(type %in% c("Company", "Beneficial Owner"))
mc3_graph_cb <- tbl_graph(nodes = mc3_nodes_cb,
edges = mc3_edges_cb,
directed = FALSE) %>%
mutate(betweenness_centrality = centrality_betweenness(),
closeness_centrality = centrality_closeness())We bin the betweenness centrality of the nodes for ease of visualization using VisNetwork later.
Due to the large data size, we filter data with betweenness > 500000 and community group < 10. Group_component is used to identify the prominent communities.
Show the code
mc3_graph_cbcb <- mc3_graph_cb %>%
activate("nodes") %>%
mutate(group = cut(betweenness_centrality, breaks = c(0, 1000000, 2000000, 3000000, Inf),
labels = c("1\n(0-999999)",
"2\n(1000000-1999999)",
"3\n(2000000-2999999)",
"4\n(>=3000000)\n"),
include.lowest = TRUE)) %>%
arrange(desc(betweenness_centrality)) %>%
filter(betweenness_centrality > 500000) %>%
mutate(group1 = group_components()) %>%
filter(group1 < 10)
mc3_graph_cbcb <- mc3_graph_cbcb %>%
activate("edges") %>%
mutate(importance = centrality_edge_betweenness()) The network graph showing the relationship between Company and Beneficial Owner is plotted.
Show the code
set.seed(1234)
ggraph(mc3_graph_cbcb,
layout = "stress") +
geom_edge_link(aes(colour = importance),
alpha=0.2) +
scale_edge_width(range = c(0.1, 5)) +
geom_node_point(aes(size = betweenness_centrality, colour = factor(group1))) +
theme_graph() +
geom_node_text(aes(label = id), size = 1, repel=TRUE) +
ggtitle("<span style='font-size: 15pt;'>Betweenness Centrality Network (Company-Beneficial Owner)</font>") +
theme(plot.title = element_markdown())
Show the code
kable(head(mc3_graph_cbcb %>%
activate("nodes") %>%
as_tibble(), 5))| id | type | country | revenue_omu | word | betweenness_centrality | closeness_centrality | group | group1 |
|---|---|---|---|---|---|---|---|---|
| Senegal Coast Ltd. Liability Co | Company | Oceanus | NA | NA | 3399060 | 2.09e-05 | 4 | |
| (>=3000000) | 1 | |||||||
| Jessica Brown | Beneficial Owner | NULL | NA | NA | 2793321 | 1.95e-05 | 3 | |
| (2000000-2999999) | 1 | |||||||
| Ocean Observers Marine mist | Company | Puerto Sol | 39678.54 | transportation , NA , services | 2691699 | 1.86e-05 | 3 | |
| (2000000-2999999) | 1 | |||||||
| BlueTide GmbH & Co. KG | Company | Azurionix | 24730.50 | fabricated , metal , products | 2542636 | 1.86e-05 | 3 | |
| (2000000-2999999) | 1 | |||||||
| David Thomas | Beneficial Owner | NULL | NA | NA | 2500803 | 1.77e-05 | 3 | |
| (2000000-2999999) | 1 |
In the next step, we plot the interactive graph using VisNetwork to better visualize who are the beneficial owners of the non-fishing companies.
4.3.3 Analysis of betweenness centrality between Company and Beneficial Owner (VisNetwork)
Show the code
edges_cbcb_df <- mc3_graph_cbcb %>%
activate(edges) %>%
as_tibble()
nodes_cbcb_df <- mc3_graph_cbcb %>%
activate(nodes) %>%
as_tibble() %>%
rename(label = id) %>%
mutate(id=row_number()) %>%
select(id, label, group)Show the code
visNetwork(nodes_cbcb_df,
edges_cbcb_df) %>%
visIgraphLayout(layout = "layout_with_fr") %>%
visOptions(highlightNearest = TRUE,
nodesIdSelection = TRUE) %>%
visEdges(smooth = list(enabled = TRUE,
type = "curvedCW")) %>%
visLegend() %>%
visLayout(randomSeed = 123)4.4 Visualization of distribution of Company vs Company Contacts
We want to check the frequency distribution of Company vs Company Contacts.
First, select edges where source type = Company and target type = Company Contacts. Count number of duplicated target i.e. Company Contacts to find out how many companies are linked to each contact. Then plot the distribution.
Show the code
mc3_edges_cc <- mc3_edges_cleaned %>%
filter(type == "Company Contacts") %>%
add_count(target) #This adds a column (n) to the table indicating the number of companies linked to the Company Contact.
ggplot(data = mc3_edges_cc,
aes(x = n)) +
geom_bar() +
labs(title = "Distribution of Number of Companies linked to Company Contacts", x = "Number of Companies Linked", y = "Number of Company Contacts")
Show the code
kable(head(mc3_edges_cc %>%
arrange(desc(n)) %>%
as_tibble(), 10))| source | target | type | weights | n |
|---|---|---|---|---|
| Náutica del Sol Ges.m.b.H. | Angela Wood | Company Contacts | 1 | 7 |
| Náutica del Mar S.A. de C.V. Carriers | Angela Wood | Company Contacts | 1 | 7 |
| Costa del Sol Carriers | Angela Wood | Company Contacts | 1 | 7 |
| Ancla de Oro United Yacht | Angela Wood | Company Contacts | 1 | 7 |
| Playa de Oro Company | Angela Wood | Company Contacts | 1 | 7 |
| Sparrmans Swordfish Ges.m.b.H. Merchants | Angela Wood | Company Contacts | 1 | 7 |
| Gulf of Guinea Oceanography | Angela Wood | Company Contacts | 1 | 7 |
| PacificPlates S.A. de C.V. | Mr. Jason Carrillo | Company Contacts | 1 | 6 |
| Baltic Sprat Ges.m.b.H. Enterprises | Mr. Jason Carrillo | Company Contacts | 1 | 6 |
| Rufiji Delta Limited Liability Company | Mr. Jason Carrillo | Company Contacts | 1 | 6 |
We also want to find out which company has the most number of contacts.
Show the code
mc3_edges_cc1 <- mc3_edges_cleaned %>%
filter(type == "Company Contacts") %>%
add_count(source) #This adds a column (n) to the table indicating the number of contacts linked to the company.
ggplot(data = mc3_edges_cc1,
aes(x = n)) +
geom_bar() +
labs(title = "Distribution of Number of Company Contacts", x = "Number of Company Contacts", y = "Number of Companies")
Show the code
kable(head(mc3_edges_cc1 %>%
arrange(desc(n)) %>%
as_tibble(), 15))| source | target | type | weights | n |
|---|---|---|---|---|
| Aqua Aura SE Marine life | Anthony Hunter | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Heather Erickson | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Jesus Mcclure | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Leon Pittman | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Mrs. Amy Graves | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Sarah Barrett | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Shannon Snyder | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Thomas Snyder | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Jennifer Bauer | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Jillian White | Company Contacts | 1 | 11 |
| Aqua Aura SE Marine life | Leah Cruz | Company Contacts | 1 | 11 |
| Irish Mackerel S.A. de C.V. Marine biology | Angelica Gates | Company Contacts | 1 | 7 |
| Irish Mackerel S.A. de C.V. Marine biology | Mario Lee | Company Contacts | 1 | 7 |
| Irish Mackerel S.A. de C.V. Marine biology | Tara Cooper | Company Contacts | 1 | 7 |
| Irish Mackerel S.A. de C.V. Marine biology | Wendy Gardner | Company Contacts | 1 | 7 |
4.5 Visualization of relationship between Company and Company Contacts using network graph
4.5.1 Analysis of Company Contacts associated with more than 3 companies
First, we form new nodes data table by using source and target of the edges data table. This is to ensure that the nodes in nodes data tables include all the source and target values. Group_component <10 is used to identify the prominent communities.
Show the code
mc3_edges_cc4 <- mc3_edges_cc %>%
filter(n>3)
id9 <- mc3_edges_cc4 %>%
select(source) %>%
rename(id = source)
id10 <- mc3_edges_cc4 %>%
select(target) %>%
rename(id = target)
mc3_nodes_cc4 <- rbind(id9, id10) %>%
distinct() %>%
left_join(mc3_nodes_cleaned, by=c("id" = "id"),
unmatched = "drop") %>%
filter(type %in% c("Company", "Company Contacts"))
mc3_graph_cc4 <- tbl_graph(nodes = mc3_nodes_cc4,
edges = mc3_edges_cc4,
directed = FALSE)
mc3_graph_cc4 <- mc3_graph_cc4 %>%
activate("nodes") %>%
mutate(group = group_components()) %>%
filter(group < 10)
edges_cc4_df <- mc3_graph_cc4 %>%
activate(edges) %>%
as_tibble()
nodes_cc4_df <- mc3_graph_cc4 %>%
activate(nodes) %>%
as_tibble() %>%
rename(label = id) %>%
mutate(id=row_number()) %>%
select(id, label, group)Show the code
visNetwork(nodes_cc4_df,
edges_cc4_df) %>%
visIgraphLayout(layout = "layout_with_fr") %>%
visOptions(highlightNearest = TRUE,
nodesIdSelection = TRUE) %>%
visEdges(smooth = list(enabled = TRUE,
type = "curvedCW")) %>%
visLegend() %>%
visLayout(randomSeed = 123)4.5.2 Analysis of relationship between Company and Company Contacts (tidygraph)
First, we form new nodes data table by using source and target of the edges data table. This is to ensure that the nodes in nodes data tables include all the source and target values.
Show the code
id11 <- mc3_edges_cc %>%
select(source) %>%
rename(id = source)
id12 <- mc3_edges_cc %>%
select(target) %>%
rename(id = target)
mc3_nodes_cc <- rbind(id11, id12) %>%
distinct() %>%
left_join(mc3_nodes_cleaned, by=c("id" = "id"),
unmatched = "drop") %>%
filter(type %in% c("Company", "Company Contacts"))
mc3_graph_cc <- tbl_graph(nodes = mc3_nodes_cc,
edges = mc3_edges_cc,
directed = FALSE) %>%
mutate(betweenness_centrality = centrality_betweenness(),
closeness_centrality = centrality_closeness())Due to the large data size, we filter data with closeness centrality > 0 and community group < 10. Group_component is used to identify the prominent communities.
Show the code
mc3_graph_cccb <- mc3_graph_cc %>%
activate("nodes") %>%
arrange(desc(betweenness_centrality)) %>%
filter(betweenness_centrality > 0) %>%
mutate(group1 = group_components()) %>%
filter(group1 < 10)
mc3_graph_cccb <- mc3_graph_cccb %>%
activate("edges") %>%
mutate(importance = centrality_edge_betweenness()) The network graph showing the relationship between Company and Beneficial Owner is plotted.
Show the code
set.seed(1234)
ggraph(mc3_graph_cccb,
layout = "fr") +
geom_edge_link(aes(colour = importance),
alpha=0.2) +
scale_edge_width(range = c(0.1, 5)) +
geom_node_point(aes(size = betweenness_centrality, colour = factor(group1))) +
theme_graph() +
geom_node_text(aes(label = id), size = 1, repel=TRUE) +
ggtitle("<span style='font-size: 14pt;'>Betweenness Centrality Network (Company-Company Contacts)</font>") +
theme(plot.title = element_markdown())
Show the code
kable(head(mc3_graph_cccb %>%
activate("nodes") %>%
as_tibble(), 5))| id | type | country | revenue_omu | word | betweenness_centrality | closeness_centrality | group1 |
|---|---|---|---|---|---|---|---|
| Aqua Aura SE Marine life | Company | Mawazam , Rio Isla , Icarnia , Oceanus , Nalakond , Coralmarica, Alverossia , Isliandor , Talandria | NA | food , ingredients , frozen , fruits , vegetables , ready , eat , products , ready , cook , products , snacks , drinks , kitchen , accessories , seafood , aquatic , products , bone , ash , natural , calcium , phosphate , feldspar , prepared , bone , china , bodies , NA , NA , fish , seafood , products , tuna , salmon , herring , shellfish , groundfish , products , flounder , fillets , cornmeal , pollock , strips , burger , tuna , steak , frozen , halibut , steaks , canned , sockeye , salmon , frozen , sockeye , crabs , alaskan , seafood , domestic , caviar , imported , caviar , NA , NA , NA , optical , fiber , communication , passive , components , fiber , optical , attenuator , fiber , optical , isolator , fiber , optical , switch , optical , filter , fiber , optical , fiber , optical , coupler , NA | 108.5 | 0.0357143 | 4 |
| Irish Mackerel S.A. de C.V. Marine biology | Company | Oceanus | NA | transportation , services , ceramic , resin , home , garden , decor , source , freelance , researcher , involved , retailing , fresh , frozen , cured , meats , poultry , NA , NA , NA , NA , NA , NA , NA , NA , NA | 70.5 | 0.0277778 | 4 |
| Jillian White | Company Contacts | NULL | NA | NA | 30.0 | 0.0312500 | 4 |
| Leah Cruz | Company Contacts | NULL | NA | NA | 30.0 | 0.0312500 | 4 |
| Mar del Norte NV | Company | Oceanus, Marebak | 22966.67 | seafoods , fish , NA , seafood , products , NA | 21.0 | 0.0769231 | 5 |
5. Conclusion
In conclusion, majority of the Beneficial Owners own only 1 company, which is normal. However, a minority of them own more than 4 companies, which is suspicious. For example, John Smith owns the most number of companies (11). We can only ascertain that one of the companies (Adams Group) is related to fishing activities. Beachcombers Nautical Plc Carriers is into products, mail, houses activities and it has an exceptionally high revenue_omu of 9891666.673. There is no info for the rest of the companies.
It is also observed that majority of the Company Contacts are associated with only 1 company, which is normal. However, a minority of them are associated with 4 or more companies, which is rather suspicious. For example, Angela Wood has contacts with the most number of companies (7). However, there is no info on revenue_omu nor product services for all the 7 companies other than the info that Náutica del Sol Ges.m.b.H. is involved in industrial adhesives business. It is also suspicious that Aqua Aura SE Marine life has so many Company Contacts (11).